autoconference

What problem does it solve?

It coordinates multiple autonomous researchers into structured, round-based collaboration and adversarial peer review so you can reliably synthesize better results than a single agent run.

Core Features & Use Cases

• Parallel autoresearch with conference rounds: Spawns N researcher agents that iteratively research, evaluate, and log results across independent research, poster session, peer review, and knowledge transfer phases.
• Adversarial validation and synthesis: Uses a reviewer step to challenge claims, validate or overturn findings, and then transfers only validated knowledge into shared state.
• Convergence and crash recovery guardrails: Detects convergence or budget/stall stop conditions and supports recovery via an event log for interrupted conferences.

Quick Start

Create a conference.md that defines your Goal, Mode, Success Metric or Success Criteria, and search space, then ask the autoconference skill to run the conference using that file for your desired researcher count and strategy partitioning.